Agent Memory
open-sourceSummary
Agent Memory is an open-source graph-native memory system for AI agents that stores conversations, builds knowledge graphs, and allows agents to learn from their own reasoning, all backed by Neo4j. It is licensed under Apache 2.0 and operates as a library supporting Python and TypeScript bindings, positioning it alongside LangChain due to its framework integrations. The system is intended for platform engineers or teams building agentic workflows, and its documentation shows integrations with tools like CrewAI and Pydantic AI. It has had two releases tracked within our monitoring period.
What Agent Memory answers
What types of historical data can it store?
conversations, entities, facts, and reasoning traces
Does it perform entity recognition and data enrichment?
multi-stage entity extraction using tools like spaCy/GLiNER/LLM, relationship extraction, and background enrichment from sources like Wikipedia
Which types of queries does the underlying system support?
vector search, text search, and geospatial queries
Does it connect to external development frameworks?
it has documented integrations with LangChain, Pydantic AI, Google ADK, Strands, and CrewAI
What is the mechanism for learning from past operations?
it builds a knowledge graph from conversations and reasoning traces to allow learning from past decisions
What form does the memory system take?
it operates as a library with bindings for Python and TypeScript
Examples
Command line
No option matches that search.
| option | found in | since | description |
|---|
No option matches that search.
Values are placeholders taken from each option’s declared default. Nothing is executed here — the output shown is a recording of a run that already happened.
Release history
Nothing shipped since we started watching. Releases and docs updates land here as they happen.